A Decentralized, Communication-Free Force Distribution Method With Application to Collective Object Manipulation

Abstract

We present a novel approach to achieve decentralized distribution of forces in a multirobot system. In this approach, each robot in the group relies on the behavior of a cooperative virtual teammate that is defined independent of the population and formation of the real team. Consequently, such formulation eliminates the need for interagent communications or leader–follower architectures. In particular, effectiveness of the method is studied in a collective manipulation problem where the objective is to control the position and orientation of a body in time. To experimentally validate the performance of the proposed method, a new swarm agent, Δρ (Delta-Rho), is introduced. A multirobot system, consisting of five Δρ agents, is then utilized as the experimental setup. The obtained results are also compared with a norm-optimal centralized controller by quantitative metrics. Experimental results prove the performance of the algorithm in different tested scenarios and demonstrate a scalable, versatile, and robust system-level behavior.